1
\$\begingroup\$

We have been trying to figure out for weeks why these flyback transformers keep failing. They are these specific transformers http://www.pca.com/Datasheets/EPC3205G-X-LF.pdf

We are running them at 24V input, and are used to charge a capacitor bank to 200+ Volts. We run them at around 900 mA at 355kHz in a DC/DC circuit.

I know that's not a lot to go one but we have somewhat ruled out overheating and over current as the culprit. They don't get hotter than 60 degrees C, and verified with a scope that the current does not exceed 1 A. Nearly out of ideas as to what to do so any advice would be welcome.

SOLVED:

Advice to anyone looking to build something similar:

Don't go with the PCA transformers! We ordered some nearly identical transformers from PULSE PA0367ANLT and they are doing a lot better! No more random failures. Plus they are cheaper.

\$\endgroup\$
  • \$\begingroup\$ the data sheet gives peak operation current of 1.5A at 25C so say if you are operating at 50C the maximum current rating may reduce drastically ,how ever i don't know exact value but for sure it will reduce \$\endgroup\$ – Lokanath Sep 18 '15 at 5:43
  • \$\begingroup\$ In the Electrical Parameters table the voltage input of the flyback transformer shall be between 2.5 to 5V... \$\endgroup\$ – R Djorane Sep 18 '15 at 7:57
  • \$\begingroup\$ What controller chip are you using? Since your 24-volt input exceeds the ratings of the LT3420 controller that the transformer datasheet suggests, I assume that you are either using a different controller or have adapted the LT3420 in some way. \$\endgroup\$ – Fred Schleifer Sep 18 '15 at 18:27
  • \$\begingroup\$ We are using the LT3757EDD \$\endgroup\$ – THEMuffinMan7 Sep 21 '15 at 16:15
  • \$\begingroup\$ Another thing is we are using two of these transformers to generate a dual supply. Ie the two primary windings are in parallel. Not sure if this could be an issue. \$\endgroup\$ – THEMuffinMan7 Sep 21 '15 at 16:18
1
\$\begingroup\$

Another possibility (in addition to those in the previous answer and comments) is that the pins of the secondary connection are swapped -- or equivalently, the pins of the primary connection are swapped.

If the connection to one of the windings has a flipped polarity, this circuit will work as a forward converter instead of as a flyback converter. Given the turns ratio of the transformer (1:10.2 or 1:12) and the 24-volt supply, the circuit can still charge the capacitor bank to 200 volts. But unlike a flyback converter, a forward converter has no intrinsic current limiting, so the peak transformer currents will be very high while charging the capacitors, which can damage the transformer (and other components).

In a flyback configuration, the output diode should be off when the transformer-driving transistor is on. The on time of the transistor charges the transformer with a current ramp limited by the primary inductance. When the transistor turns off, the diode conducts and the magnetic energy stored in the transformer transfers to the output capacitor. Because of the controlled energy transfer, a flyback converter can efficiently charge a capacitor starting at zero volts.

In a forward-converter configuration, both the transistor and diode conduct at the same time. The only things that limit the current are circuit and winding resistances, plus a very small amount of transformer leakage inductance. Not only are the peak currents very high, but directly charging a capacitor from zero volts with a forward converter dissipates half the energy into the circuit resistances. This results in a best-case efficiency of only 50 percent. If efficiency is not a concern, it's okay to use a forward converter to charge capacitors, but the design should include a resistor in the primary or secondary circuit to limit the peak charging current.

One thing that contradicts this hypothesis is your measurement that shows a maximum current of 1 amp. Was this a measurement of the transformer primary current or of the average input current to the circuit? Was the capacitor bank already partially charged when this measurement was taken?

\$\endgroup\$
  • \$\begingroup\$ We saw about 1 amp going across the current sense reisistor of the primary winding (on the scope) \$\endgroup\$ – THEMuffinMan7 Sep 21 '15 at 16:19
1
\$\begingroup\$

At 355kHz and (let's say) 50:50 duty cycle, the MOSFET will connect the transformer primary to 24V for a period of 1.41 us. The data sheet linked covers inductors from 10uH to 25uH so, assuming 25uH, the question is "what has the current in the primary risen to after 1.41us has elapsed": -

V = \$L\dfrac{di}{dt}\$ so \$di = V.\dfrac{dt}{L}\$ = 24 x 1.41us / 25uH = 1.354A

The data sheet says the maximum primary current of the 25uH model is 1.4 amps so, if your duty cycle goes a little above 50:50, then you are exceeding the primary current rating and the core will start to badly saturate and the device will certainly fail after a short period because saturation causes more current to be taken.

If it's a 10uH primary inductance then the peak current will be 2.5 x higher at 3.384 amps and both of the two 10uH transformers in the data sheet will be seriously affected because the best one has a max primary current of 1.4 amps.

However, the LT3420 will come to the rescue to a certain extent - it nominally current limits the primary current to 1.4 amps (20mV internal reference and 0.015 ohm internal measurement resistor) but, it could be a maximum of 1.6 amps and at this level I feel your design is flawed.

I feel you are asking too much of this transformer.

\$\endgroup\$
  • \$\begingroup\$ @THEMuffinMan7 do you have any questions about my answer because it would be nice to close this down with a formal acceptance. You do appear to be agreeing with me in your question editr about the transformer originally tried. \$\endgroup\$ – Andy aka Aug 22 '17 at 9:20
0
\$\begingroup\$

Advice to anyone looking to build something similar:

Don't go with the PCA transformers! We ordered some nearly identical transformers from PULSE PA0367ANLT and they are doing a lot better! No more random failures. Plus they are cheaper.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.